Research On Virtual Machine Resource Scheduling Methods For Cloud Service

As a new type of network service delivery model,cloud services realize the reception and processing of user service requests based on cloud computing technology and Internet information transmission technology.The service model breaks through the hardware limitations of the terminal device and expands the computing,storage,and load balancing capabilities of the terminal device.In addition,with the rapidly increasing number of network users,how to realize the rational management of cloud resources has become an important direction for the development of cloud services.Establishing a secure and credible cloud service mechanism has also become a research hotspot in this field.In the trusted three-tier management mechanism "user-environment-service",service trust is the core issue to ensure cloud service provision.In order to realize the trustworthiness of resource management of cloud services,this thesis proposes a resource management scheme from the standardization management of service requests and the random application scheduling of resources,which can realize the scheduling of virtual machine resources according to service requests and changes of cloud resources and application of models.Resource management scheme also can use the model's generation strategy to achieve resource pre-adjustment.The specific research work includes the following two aspects:1.Aiming at the complexity of service requests in cloud services and the lag of service resources,a service-aware resource management framework is proposed.The framework performs matching of service requests and virtual machines according to different types of services,resource requirements,and virtual machine service capabilities,and implements management of virtual machine resources.On this basis,the request classification algorithm and the virtual machine power on/off management algorithm are proposed to cooperate with the framework for performing request classification and virtual machine adjustment.In addition,the proposed virtual machine migration algorithm can guide the virtual machine migration according to the change of the service request arrival rate,realize the pre-adjustment of the corresponding service resources,reduce the service lag,and improve the service efficiency.2.In order to realize the dynamic and adaptive nature of resource management,and fully considering the location characteristics of users in cloud services and the complexity of network structure,a stochastic resource scheduling method is proposed.This method uses the Markov decision process to simulate the relationship between service requests and random scheduling.The optimization goal is then formalized into an optimization problem to solve and guide the resource scheduling of the entire service.The method can also dynamically adjust the scheduling policy according to the change of the service request arrival rate,thereby realizing pre-scheduling of resources,enhancing resource dynamics and adaptive management.Finally,experiments show that the two resource scheduling methods proposed in this thesis achieve the goal of improving resource utilization,reducing service delay and cost while realizing resource rationalization scheduling.The research in this thesis also provides some useful references for the research of trusted service resource management in the cloud environment.